At MWC 2026 in Barcelona, the GSMA, the lobbying organisation for the mobile communications industry, launched Open Telco AI, a new industry initiative aimed at building AI models that actually work for telecom use cases. According to the GSMA, Open Telco AI is “a new portal designed to bring operators, vendors, researchers, and developers together with shared resources, datasets, tools, and benchmarking.”
Here’s what that means and why it matters.
Why Telecom AI Hasn’t Worked Well So Far
According to the GSMA, most frontier AI models still struggle with telecom-specific tasks. That includes:
- Interpreting complex network data
- Understanding telecom standards documentation
- Automating network operations with sufficient accuracy
GSMA Intelligence data shows only 16% of GenAI deployments in telecom are being used in network operations, highlighting the gap between AI hype and operational impact.
As GSMA’s Director of AI Technologies, Louis Powell, put it, “AI does not yet speak telco and operators are often deploying technology that cannot meet the required levels of accuracy, safety or efficiency. Establishing clear benchmarks and collaborating across the industry on datasets, models and agentic systems is essential.”
What Open Telco AI Is Offering
Open Telco AI is positioned as a shared ecosystem rather than a single product.
Through GSMA.com/open-telco-ai, the initiative provides:
- Open telco-specific AI models
- Datasets tailored to telecom tasks
- Compute resources
- Benchmarking tools
- A performance leaderboard
Progress will be tracked through a new Telco Capability Index, which measures how well AI models perform across telecom-specific tasks.
For operators and vendors, this creates a clearer way to evaluate whether AI tools are actually production-ready.
Who’s Backing It
The initiative launches with support from:
- AT&T — releasing open telco-trained models built on publicly available data, designed to be hardware- and cloud-agnostic
- AMD — providing GPU compute for training, fine-tuning, inference, and evaluation
- TensorWave — hosting the infrastructure
Why This Matters for Operators
If successful, Open Telco AI could help telecom companies:
- Reduce reliance on generic AI models that underperform in network environments
- Benchmark model quality before deployment
- Fine-tune models specifically for telecom workflows
- Collaborate across operators, vendors, and researchers
In short, it aims to move telecom AI from experimentation to operational deployment.
The Bigger Picture
Telecom has lagged behind other sectors in applying generative AI to core operations. Open Telco AI represents an attempt to close that gap through open collaboration, shared compute, and standardized benchmarking.
Whether it works will depend on adoption. But for telecom operators looking for AI that understands their world, this is one of the first coordinated attempts to build it.
Damilare Odedina
